Evaluation of a commercial intravaginal thermometer to predict calving in a Hungarian Holstein-Friesian dairy farm

Accuracy to Predict the Onset of Calving in Dairy Farms by Using Different Precision Livestock Farming Devices

Simple Summary

If the onset of calving can be accurately detected as well as appropriate calving assistance can be performed on a dairy farm, at that time, the prevalence of dystocia, stillbirth, vaginal laceration, retained fetal membranes, and consequent clinical metritis/endometritis can be decreased significantly. Therefore, in order to reduce these losses, our primary task must be to predict the onset of calving accurately and provide timely and professional calving assistance. This review focuses on the diagnostic possibilities and limitations of detecting the onset calving in the field.

Abstract

Besides traditional methods such as evaluation of the external preparatory and behavioral signs, which even presently are widely used also in large dairy farms, there are several new possibilities such as measuring body (intravaginal, ventral tail-base surface, ear surface, or reticulo-ruminal) temperature, detecting behavioral signs (rumination, eating, activity, tail raising) or detecting the expulsion of the device inserted into the vagina or fixed to the skin of the vulva when allantochorion appears in the vulva to predict the onset of the second stage of calving. Presently none of the single sensors or a combination of sensors can predict the onset of calving with acceptable accuracy. At the same time, with the exception of the iVET^® birth monitoring system, not only the imminent onset of calving could be predicted with high accuracy, but a significantly lower prevalence rate of dystocia, stillbirth, retained fetal membranes, uterine diseases/clinical metritis could be reached while calving-to-conception interval was significantly shorter compared with the control groups. These results may confirm the use of these devices in dairy farms by allowing appropriate intervention during calving when needed. In this way, we can reduce the negative effect of dystocia on calves and their dams and improve their welfare.

Multicow pose estimation based on keypoint extraction

Automatic estimation of the poses of dairy cows over a long period can provide relevant information regarding their status and well-being in precision farming. Due to appearance similarity, cow pose estimation is challenging. To monitor the health of dairy cows in actual farm environments, a multicow pose estimation algorithm was proposed in this study. First, a monitoring system was established at a dairy cow breeding site, and 175 surveillance videos of 10 different cows were used as raw data to construct object detection and pose estimation data sets. To achieve the detection of multiple cows, the You Only Look Once (YOLO)v4 model based on CSPDarkNet53 was built and fine-tuned to output the bounding box for further pose estimation. On the test set of 400 images including single and multiple cows throughout the whole day, the average precision (AP) reached 94.58%. Second, the keypoint heatmaps and part affinity field (PAF) were extracted to match the keypoints of the same cow based on the real-time multiperson 2D pose detection model. To verify the performance of the algorithm, 200 single-object images and 200 dual-object images with occlusions were tested under different light conditions. The test results showed that the AP of leg keypoints was the highest, reaching 91.6%, regardless of day or night and single cows or double cows. This was followed by the AP values of the back, neck and head, sequentially. The AP of single cow pose estimation was 85% during the day and 78.1% at night, compared to double cows with occlusion, for which the values were 74.3% and 71.6%, respectively. The keypoint detection rate decreased when the occlusion was severe. However, in actual cow breeding sites, cows are seldom strongly occluded. Finally, a pose classification network was built to estimate the three typical poses (standing, walking and lying) of cows based on the extracted cow skeleton in the bounding box, achieving precision of 91.67%, 92.97% and 99.23%, respectively. The results showed that the algorithm proposed in this study exhibited a relatively high detection rate. Therefore, the proposed method can provide a theoretical reference for animal pose estimation in large-scale precision livestock farming.

A field study using different technologies to detect calving at a large-scale hungarian dairy farm

The objective of this study was to investigate three different calving detection systems in order to assess and compare their efficiency. The study was conducted at a large-scale dairy farm involving 54 Holstein-Friesian dairy cows and heifers. Animals were fitted with multiple devices: a rumination measuring device (Ruminact^® (RA)), an intravaginal thermometer (Vel'Phone^® (VP)) and a tail movement sensor (Moocall^® (MC)) 5 to 7 days before expected calving and were removed after parturition. RA detects rumination time (RT) and calculates it in 2-hr intervals. VP detects a decrease in vaginal temperature that might indicate calving within 48 hr and the drop in temperature resulting from the expulsion of the device at calving (EXP message). MC detected increased tail movements and if they persisted for one hour, 1HA message was sent. If they continued during the subsequent hour, then 2HA message was sent. Messages sent by MC within 4 hr before calving (C4) were selected retrospectively as true positives for the prediction of calving, using the significant changes in RT as a baseline. All other messages were categorized as false positive. The mean value of RT decreased in a non-significant manner between interval -22 and -4 before calving. Significant decrease of RT was detectable between the two intervals of -4 and -2 before calving (24.7 ± 18.6 min/2 hr and 14.0 ± 13.0 min/2 hr, respectively). There was no significant difference between RT of primiparous and multiparous animals. EXP messages were accurate (positive predictive value 100%) indicators of the onset of calving. We received on average 12.7 ± 15.2 messages/animal (11.0 ± 10.1 and 16.6 ± 22.2 for cows and heifers, respectively). Positive predictive value was 12.6%. The number of false-positive messages was significantly higher in heifers. All three automatic systems could be used in a large-scale farm environment.